Traffic light control is the most important and efficient method for controlling traffic in urban areas. There are three categories of traffic light control strategies: fixed-time control, traffic actuated control and traffic adaptive control. In fixed time control each traffic light has a predefined duration for allowing traffic to flow. The controller cycles between all the traffic signals. In this manner each lane gets a predefined duration of a green light and flow of traffic. The predefined green light duration is determined by the system designer based on expected traffic flow; relative road size; time of day; and other considerations. As the rate of flow of traffic increases, the fixed time control may not provide the optimal division of time between the different lanes and traffic congestion may arise.
A remedy to the inefficiencies of fixed time traffic light control is to measure the traffic flow and change the traffic light duration according to measured traffic flow. Examples of measuring traffic flow include; wire loops embedded in the road which generates a current when a car passes over them; pressure sensitive devices embedded in the road; acoustic devices to measure traffic flow; and image based systems to measure traffic flow. Examples of existing algorithms are given in Ref [CYBERNETICS AND INFORMATION TECHNOLOGIES, Volume 13, No 3 DOI: 10.2478/cait-2013-0029] and Ref[Self-Algorithm Traffic Light Controllers for Heavily Congested Urban Route, WSEAS TRANSACTIONS on CIRCUITS and SYSTEMS, Issue 4, Volume 11, April 2012.
Existing solutions focus on the static measure of traffic. For example, a green traffic light is provided if the traffic sensor indicates the existence of a car in the relevant lane. To assess the amount of traffic in the lane these solutions require additional sensors which increases the cost of deployment and significantly increases the cost of operation. The required computing resources for some algorithms are not supported by existing traffic light controllers, so a deployment of some systems requires an overhaul of the existing infrastructure. Furthermore, the cited examples which dynamically change the traffic light duration based on static measurements also change the cycle of the traffic signal. Changing the cycle of the traffic light disrupts the flow of traffic and induces congestion across the road system. In some examples of state of art solutions, the applied methods can only reduce the preplanned maximum time for each light. In one example, if 20 seconds are allocated for a green light than the state of art method will reduce the allocated time from 20 seconds to a smaller number. Hence this will shorten the allocated green time to a specific lane, without an increase in green light time to other lanes. The inability to increase the allocated green light time to more than 20 seconds results in traffic congestion as demonstrated in this example. Assuming traffic is congested and requires 23 seconds to pass through the junction. If only 20 seconds are allocated than the remaining 3 seconds of traffic would be stopped for next green light cycle. In the next green light cycle, there will now be 23 seconds of traffic plus the 3 seconds from the previous cycle. Thus the traffic flow is impeded and congestion arises rapidly.
Hence an alternative algorithm is required. The desired algorithm should provide the following features:
Can be implemented in the existing infrastructure of controllers and single sensor per lane.
Maintains the traffic flow cycle to prevent disruption to traffic.
In some examples, the proposed algorithm can also increase the allocated green time beyond the static allocated green time.